I didn't dive to the code at this point however I remember that in the past they did use a video for 3D animations. And scrolling was just playing the video timeline forward and backward.
A quick peek through the inspector yesterday showed me that they (for the part in the gif) had all three phones as three different elements that were getting moved around.
Binding positions to scroll-offsets, using text, videos and transparent images mostly.
The real challenge is not the how, but the what. Web/Motion-Designers, developers, marketing need to work closely together for a good result. Apple is obviously good at this.
We're a place where coders share, stay up-to-date and grow their careers.
We strive for transparency and don't collect excess data.